Everybody knows that water is a precious good. Governments and citizens make great efforts to store, purify and supply water and there are thousands of miles of large-diameter pipelines that carry water over long distances. In the vicinity of cities, these pipelines branch out into supply networks. The conservation of such infrastructures is not always adequate, resulting in great water losses, in some cases as high as between 40% and 60% of the water transported.
There are methods to inspect supply pipes but efficient methods for the analysis of pressurised, large-diameter pipelines do not exist.
We are all aware of the problem of successfully locating water leaks in large-diameter networks. These problems have their roots in the physics principle used by traditional technologies (loggers, correlators, geophones) to detect such leaks. These electronic devices amplify the sound caused by the friction of the water leaving the pipe at the point of leakage.
They are highly effective in supply pipes because, due to the structure of these networks, the sound caused by the leak can “travel” through the pipe wall for many metres. Induction occurs of the sound to the numerous elements of which these distribution networks are composed and these technologies carry out much of the work involved in capturing sound.
Indeed, the sound transmitted in these networks even reaches the surface through the soil.
The inspection of underground pipes using traditional methods is carried out from the surface, meaning that extraneous sounds, such as traffic noise, can “overlap” with the sound caused by water leaks.
All these traditional systems are ineffective in large-diameter networks. These networks tend to be deeper, with fewer elements and greater distances between them, thereby hindering that task of recording sounds. Moreover, the sound originating from the leak becomes diluted in a matter of very few metres, due to the structure and volume of a large-diameter pipeline. All this makes it practically impossible to locate leaks successfully with traditional systems.
For this reason a working system (Nautilus) has been patented and developed. This system is currently being commercialised in the form of inspection services and licenses for geographic exclusivity.
This article presents the contribution made by the Nautilus working system and the results obtained in field tests, along with some conclusions and future lines of work.
Nautilus is a leak detection system for pressurised large-diameter pipelines, which carries out unattended inspection of kilometres of pipeline per day.
The Nautilus working system consists of a sphere of minimum dimensions (86 mm), which is inserted into the water distribution system and is carried internally by the water. The sound created by an anomaly arising from a leak or air pocket is characteristic and familiar. The sphere monitors sounds inside the pipeline. When it is extracted, the data collected is sent to a software program for processing by means of a mathematical algorithm and the location of such anomalies is indicated by means of GPS.
The operating principle of the Nautilus system is based on the estimation of the position of the system as it travels through the pipeline, in accordance with the speed of the water flow and the time that has passed since insertion. A single propagation direction, the axial direction, is assumed. Because the exact point of anomalies is unknown, Kalman filtering is applied to identify the non-measurable points of the route.
The Nautilus system process is as follows:
The sphere is inserted into a suction cup of at least 100 mm and is capable of passing through butterfly valves. The diameter of the inside of the pipeline to be analysed ranges from 400 mm to 1600 mm and the material of which the pipeline is built can be parametrized. Inspections can be carried out continuously over 36 hours. Synchronisers must be placed at certain intervals, which minimises position error, with the disadvantage being the time needed to put the synchroniser in place.
Flow speed inside the pipeline is neither constant nor known. For this reason, several measurements are taken to estimate the speed, and correction by means of Kalman filtering is applied. The synchronisers placed in known positions emit a characteristic sound, which is monitored by Nautilus. Because the position of this synchroniser is known a priori, the error is determined and corrected. The synchronisers are positioned at access points of the central pipeline, with the ideal points being those closest to the inside.
The sphere travels through the inside of the pipeline until it reaches a tank. If there is no tank in the section of pipeline being analysed, it is extracted by means of an access suction cup of 150 mm, using a net, which is inserted at pressure and withstands pressures of up to 25 atm. A minimum water speed is required for the sphere to move forward towards the net but the water speed cannot be so high as to cause the sphere to go through the net. The measurable values are between 0.4 and 1.6 m/s.
Because there are known values and predicted values, which are different sounds, cubic spline interpolation is applied to smoothen out the changes created by the different speeds in different sections. The sound analysed can contain different spurious noises, which are filtered using different signal processing techniques (Hamming Window, spurious noise Filter, Butterworth, Chevyshev, Bessel Filter). To minimise error in calculations, the pipeline is divided into sections, with each section being the distance between synchronisers.
In the preliminary simulations, ideal conditions, without external noises and with constant speed and flow were assumed. In a real inspection, different spurious noises are generated and the speed of the section being inspected is not constant, because in a pressurised pipeline, the different water demands of the branch lines change the flow in the main channel. A test inspection was undertaken to find the typical values of the Nautilus working system.
In the inspection to be analysed, the pipe had a diameter of 800 mm and was made of steel-jacketed concrete. The inside of the pipe contained butterfly valves and an elbow of 90º with a drop of 8 m, a horizontal section of 40 m and a vertical rise of 8 m. Analysis of the inspection was broken down into four stages: insertion, monitoring, extraction and processing.
A 100 mm suction cup was used for insertion of the Nautilus working system. The speed of the water ranged from 0.8 and 1.2 m/s so the sphere had to be pushed towards the centre of the central pipeline.
Once inside, the Nautilus working system monitors the inside of the pipe with a 44 Khz sampling rate and stores the anomalies found, along with diverse sounds. There were no direct connections with the central pipeline so the synchronisers were positioned at the points closest to the main pipeline, with a distance of 5 meters between the branch line and the pipeline to be inspected. Because there were no intermediary points, distance between synchronisers of 1,600 m were created.
There was a tank in place to carry out extraction. A metal mesh prevented the sphere from advancing with the water.
When carrying out the processing, filtering must be undertaken to eliminate noise, increasing processing time to 15 minutes / hour of recording (Intel i5-4460 3.2GHz, 8GB RAM)
When carrying out the analysis, an anomaly with a margin of error of 7 metres from the source of the anomaly, without applying error correction, was detected in the section without synchronisers, with a spacing between synchronisers of 1600 m.
Different simplifications were assumed, such as constant water speed inside each section of pipeline. Bearing in mind that this speed is not constant, spline cubic interpolation was carried out. This corrected the margin of error of anomaly location to 2 meters.
The displacement of the audio spectrum when the sound wave changes material is worth pointing out, as is the fact that a large quantity of noises in the recording hinders automation, due to the fact that the algorithms used in the simulations did not take these noises into account.
The working system and the methodology are protected by an international patent (PTC).
The system presented is for large diameter (400 mm to 1600 mm) pipelines. It is capable of inspecting long distances in a short time (17.2 Km / day) and is based on a sphere that travels through the inside of the pipes. A minimum water speed of 0.4 m/s is required for correct displacement of the sphere. Inspection is carried out with the pressurised pipeline in operation and pressures of up to 25 atm can be withstood. Synchronisers placed at intervals of 500 m are needed to minimise errors in locating the anomalies found inside the pipelines. Using this method, leaks of 0.04 l/s have been located.
Ongoing development is now focusing on reducing the size of the sphere, minimising location errors and integrating the sensors to check the status of the water.
This work was undertaken with the support of the Department of the Environment and Territorial Planning of the Government of Andalusia, who cooperated with us in the research, testing and calibration of the system.